Photography Summary

本文介绍了夜景摄影的七大技巧,包括抓准黄金时间、使用三脚架和快门线、选用RAW格式拍摄、善选机位、尝试创意玩法、合理运用大光圈和小光圈以及正确测光等。这些技巧有助于提升夜景摄影作品的质量。

一:抓准黄金时间。
夜幕将至未至的时间拍摄夜景的。此时,华灯初上,晚霞刚隐,天空的颜色不是黑,而是带着一抹淡淡的彩霞余光,而此时城市的夜生活已然开始,人头涌动,好不热闹。画面的色彩层次感强,色彩华丽,但城市的夜景也勾勒得十分恰当,十分漂亮。些许的云彩和晚霞的余光。

二:三脚架,快门线,低ISO。
这个不用多说了,拍夜景你一般都需要好的机位和长曝光,所以这两样东西是最基本的了。而在这两样东西的支撑下,你就没有必要用高ISO了,毕竟低ISO的质量更好。

三:用RAW。

四:善选机位:天桥 江边 人行道

五:创意玩法:光绘,变焦 追焦

六:大光圈,小光圈。
大光圈可以拥有梦幻般的的焦外,小光圈可以获得更久的曝光时间和星星般的点光,根据你的需要选择光圈。

在夜景拍摄中,绝大多数场景推荐使用A挡光圈优先,因为可以随意控制景深。
比如拍风光车流什么的就小光圈,低iso,拍妹子或者炫耀焦外就大光圈(iso由希望的快门速度决定)。

七、测光
然后测光呢?这个要看拍摄场景选。
如果拍摄场景中只有主体一部分是亮的,就对它进行点测光。
想象北京的天际线上有一座高楼,很亮,天空漆黑,就用点测光测高楼,(可能要测光后重新构图)然后曝光就对了。你要是点测黑暗的天空,那相机吧天空曝光成正常的,那高楼早白的找不到了。
要是场景中有好多都是亮的,就用平均测光和中央重点测光就好。

现在有一个Jsonl文件,有id,image,width,length,conversations等字段,其中一条数据示例如下:{“id”: 1276297, “image”: “sharegpt4v_mix665k_cap23k_coco-ap9k_lcs3k_sam9k_div2k/000000302685.jpg”, “width”: 640, “height”: 427, “conversations”: [{“from”: “human”, “value”: “\nWhat are the two men wearing in the image?”}, {“from”: “gpt”, “value”: “The two men in the image are wearing coats, ties, and hats. This indicates a professional or formal appearance.”}, {“from”: “human”, “value”: “Where are the men standing in relation to the truck?”}, {“from”: “gpt”, “value”: “The men are standing next to the truck, indicating that they are either working with or associated with the truck in some way.”}, {“from”: “human”, “value”: “Is the photo in color or black and white?”}, {“from”: “gpt”, “value”: “The photo is an old black and white photo, which might suggest that the image was taken some time ago when color photography was not as prevalent.”}, {“from”: “human”, “value”: “What can be concluded about the men’s relationship to the truck?”}, {“from”: “gpt”, “value”: “Based on the image, it can be concluded that the two men are likely to be associated with the truck in a professional capacity. Their attire, which includes coats, ties, and hats, suggests that they could be the owners or operators of the truck, or possibly be involved in a business related to the truck. \n\nThe fact that they are standing next to the truck implies that they want to be identified with it or show it as part of their work environment. This could mean that the truck may serve a specific purpose within their profession or be part of their daily work-related activities. It is also possible that the truck may be a work vehicle designed to transport goods or perform a certain function that is relevant to their occupation.\n\nIn summary, the men’s professional attire and their proximity to the truck in the image suggest a strong connection between them and the vehicle, likely related to their work or business.”}]}针对每条数据要生成{ "modality" : "image", "tasks":{ "VQA":{ "conversations":data.get('conversations',[]) }, "classification" : "null", "depth_estimation": "null", "segmentation": "null", "detection": "null" } }这种格式的json文件,每条数据输出一个json文件,给出python代码
07-04
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值